Noise-Immune ECG Classifier Using Wavelet Transform and Neural Networks

Abstract

This paper proposes a novel algorithm for automatic classification of electrocardiogram (ECG) beats recorded by Holter systems. The algorithm is based on a combination of neural network and discrete wavelet transform. Discrete wavelet transform coefficients are used as an input of the neural network to perform the classification task. The proposed classifier wastested by both real ECG signals andartificially generated signals. Five Hermite functionswereused in generating the ECG artificial testing signals. Different levels of noise were added to the signals to examine the noise immunity of the classifier. The main advantage of the proposed classifier is that it is noise immune and accurate. The testing results on the proposed classier show that it is capable of recognising 40 beats, and it works properly in the classification of the ECG signal with a classification ratio of 100% for an SNR of more than 6 dB.